21
The Canadian Mineralogist Vol. 32, pp. 21-30 (1994)
A-SITE DISORDER IN SYNTHETIC FLUOR-EDENITE, A CRYSTAL-STRUCTURE STUDY
KATHLEEN F. BOSCHMANN, PETER C. BURNS, FRANK C. HAWTHORNE, MATI RAUDSEPP1 AND ALLAN C. TURNOCK
Depanmentof Geological Sciences, Universityof Manitoba, Winnipeg, Manitoba R3T 2N2
ABS1RACT
Large crystals of amphibole were synthesized by slow cooling of a bulk composition nominally that of end-member fluor-edenite. Electron-microprobe analysis shows the crystals to be zoned from a core composition close to end-member fluor-edenite to a rim composition close to the fluor-tremolite - fluor-pargasite join. The crystal structure of a fragment of average composition Na0 2(Ca1.926Mg0.183)(Mg4.636Al0.364)(Si6.507Al1.493)022F2, a 9.821(1), b 17.934(4), c 5.282(1) A, ,9j � 105.08(1)0, V 898.3(1) A, C21m, Z = 2, was refmed to an R index of 2.9% for 1068 observed reflections collected with MoKcx X-radiation. Tetrahedrally coordinated Al is ordered at T(l), and octahedrally coordinated Al is ordered at M(2). The Na within the A-site cavity occupies both the A(m) and A(2) sites, and site-occupancy refmement shows Na to be equally partitioned between these two sites. This ordering scheme can be explained on the basis of local bond-valence requirements of the anions coordinating the T(1) site.
Keywords: amphibole, fluor-edenite, synthesis, crystal-structure refinement, ordering, electron-microprobe analyses.
SOMMAIRE
Nous avons reussi a synthetiser de gros cristaux d'amphibole en refroidissant lentement un melange d'une composition globale egale a celle du pole fluor-edenite. Une analyse a Ia microsonde electronique demontre que ces cristaux sont zones, d'un coeur dont Ia composition est celle qui etait prevue a une bordure dont Ia composition est proche de Ia solution solide fluor-tremolite - fluor-pargasite. La structure cristalline d'un fragment dont Ia composition moyenne serait Nao.912(Ca1.926Mgo.Js3)(M�.636Alo.364)(Si6.507Al1.493)022F2, a 9.821(1), b 17.934(4), c 5.282(1) A, � 105.08(1)0, V 898.3(1) 3 A , C2/m, Z = 2, a ete affinee jusqu'a un residu R de 2.9% en utilisant 1068 reflexions observees (rayonnement MoKcx). L'aluminium a coordinence tetraedrique occupe T(1), tandis que !'aluminium a coordinence octaedrique occupe M(2). Le Na du site A occupe les sites A(m) et A(2); l'affmement indique une repartition egale de l'atome Na entre ces deux sites. Ce schema de mise en ordre decoule des exigeances locales des valences de liaisons des anions en coordinence avec T(l).
(Tradnit par Ia Redaction)
Mots-cles: amphibole, fluor-edenite, synthese, affinement de Ia structure cristalline, mise en ordre, donnees de microsonde electronique.
INTRODUCTION crystals. Na et al. (1986) reported the coexistence of diopside and sodian phlogopite up to 3 kbar within the There have been several attempts to synthesize field of edenite - richterite (- tremolite) solid solution, edenite, ideally NaCa2Mg5Si7Al022(0H)2, but none of and the results of Raudsepp et al. (1991) are broadly themhave been very successful. Widmark (1974) and compatible with this. Thus the edenite end-member Hinrichsen & Schi.irmann (1977) both reported the has not been synthesized as yet, and there is a strong synthesis of edenite, but the run yields were low, and possibility that it may not be stable. On the other hand, Greenwood (1979) concluded that Widmark (1974) fluor-edenite has been synthesized on end-member had synthesized overgrowths of tremolite on his seed composition (Graham & Navrotsky 1986). Raudsepp
'Present address: Department of Geological Sciences, University of British Columbia, Vancouver, British Columbia V6T 1Z4. 22 TilE CANADIAN MINERALOGIST et al. (1991) attempted to synthesize large crystals of fluor-edenite by slow cooling of the nominal bulk composition from above the liquidus. Although elec tron-microprobe analysis showed that the crystals depart from the ideal fluor-edenite composition, the crystals were sufficiently large to characterize by crystal-structure refinement; the results are presented here.
EXPERIMENTAL
Synthesis
Forty mg of starting materials (fused silica glass, y-Al203, CaF2, Na2C03, MgO) were sealed in a -1A-- 4 x 23 mm Pt tube and placed in a conventional verti cal quench-furnace. The run was held at 1240°C for 1 h and then cooled to 816°C over a period of 332 h. Further experimental details are given by Raudsepp et al. (1991). In general, the run products were >90% amphibole, with minor forsterite, plagioclase, clinopyroxene and cristobalite. One particular run gave one large crystal (-10 x 3 mm) with minor fine-grained non-amphibole phases; this crystal was gently crushed to give more conveuiently sized fragments, one of which was used in thepresent study.
Collection of the X-raydata
An irregular lath-like fragment of synthetic fluor edenite was mounted on a Nicolet R3m automated four-circle diffractometer. Twenty-two reflections were centered using graphite-monochromated MoKa X-radiation. The unit-cell dimensions (Table 1) were derived from the setting angles of twenty-two auto matically aligned reflectionsby least-squares methods. A total of 1435 reflections was measured over the range (3° < 29 < 60°) with index ranges 0 � h � 13, 0 � k � 25, -7 � l � 7; reflections forbidden by the C face-centering were omitted. Two standard reflections were measured every fifty reflections; no significant changes in their net intensities occurred during data
TABLE 1. MISCELLANEOUS INFORMATION FOR FLUOR-EDENITE
Space group C2/m Crystal size (mm) 0.1Ox0.24x0. 1 0 FIG. 1. Difference-Fourier sections throughthe centralA(2/m) site with the A cation removed from the structural model: a(A) 9.821(1) Total Ref. 1435 (a) (100) section at x = 0; (b) (010) section at y = \12; (c) b!Al 17.934(4) [l(obsl > 2.5a(l)) 1068 section through the electron-density maxima on the 2-fold c(A) 5.282(1) axis and in the mirror plane, approximately parallel to Pl0l 105.08(1) Final Rlobs) 2.9% (201).In (a) and (b), the contour interval is 1 e/A3• In (c), VIA3l 898.3(1) Final wR!obs) 3.0% the contour interval is 1 e/A3 below 6 e!A3, and 0.1 e!A3 above 6 e/M; only this close contouring at high levels of Ideal cell contents: 2(NaCsaMg,Si,AIO,.F2l density brings out the separate maxima on the 2-fold axis cell contents: Analyzed 2!Na..812Cs,..,.Mo..so1Aio.-1Si�..,AI1,492)022F21 and within the mirror plane, and emphasizes the minimum R = IIIFoi·IFcl}f IIFol at thecentral (0 \12 0) position;this minimum is dot-shaded wR = !Iw!JFoJ·JFcJl'IIwf'!I"'· w=1 todistinguish it fromthe surrounding maxima. A-SITE DISORDER IN FLUOR-EDENITE 23 collection. An empirical absorption-correction based TABLE 2. ANAL POSITIONAL PARAMifrERS AND EQUIVALENT ISOTROPIC DISPLACEMENi PARAMETERS on 396 psi-scan intensities was applied, reducing R (.A2x104) FOR FLUOR-EDENITI: (azimuthal) from 1.3 to 0.9%. The data were corrected for Lorentz, polarization and background effects, and X y 11 u..,... reduced to structure factors; of the 1435 reflections n1l 0.28176(8) 0.08479(4) 0.3025(2) 82(2) measured, there were 1068 unique observed [I � 2.5o(I)] reflections. n2l 0.29057(8) 0.17251(4) 0.8113(2) 71(2} M(1) 0 0.08902(7) 1/2 66(4)
Structure refinement M(2) 0 0.17547(7) 0 64(4)
M(3) 0 0 0 66(6) Scattering curves for neutral atoms, together with M(4) 0 0.27854(5) 1/2 115(3) anomalous dispersion corrections, were taken from Cromer & Mann (1968) and Cromer & Liberman A(m)** 0.0415(9) 1/2 0.0953(20) 200* (1970), respectively. The Siemens SHELXTL PLUS A(2)** 0 0.4765(5) 0 200*
(PC Version) system of programs was used throughout 0(1) 0.1084(2) 0.0860(1) 0.2182(4} 97(5) this study. R indices are of the form given in Table 1, 0(2) 0.1191(2) 0.1714(1) 0.7312(4) 91(5) and are expressed as percentages. Least-squares refinement of the structure in the 0(3) 0.1044(3) 0 0.7133(5} 107(7) space group C2/m, with starting parameters taken from 0(4) 0.3656(2) 0.2504(1) 0.7884(4) 115(6) Hawthorne (1983) and an isotropic displacement 0(5) 0.3511 (2) 0.1391(1) 0.1108(4) 128(6) model, converged to an R index of 5.1%. Conversion 0(6) 0.3462(2) 0.1164(1) 0.6075(4) 133(6) to an anisotropic displacement model, together with the refinement of all parameters, converged to a 0(7) 0.3448(3) 0 0.2784(7) 129(8) wR R index of 3.2% and a index of 3.3%. A model . with weighted structure-factors and an isotropic fixed during refinement • * A(ml and A(2) populations are 0.24(1 land 0.25(1l Na, extinction correction was tried but did not lead to any respectively. improvement in the refinement. As is usual for amphi boles with a fullA -site, the A(2/m) displacement factor was found to be very large (U "' 0.16), indicative of significant positional disorder in the cavity sur TABLE 3. ANISOTROPIC DISPLACEMENT PARAMETERS CA"x104) rounding theA site (Hawthorne & Grundy 1972). This FOR FLUOR-EDENITE feature is of considerable interest in the present case u, u22 u.. u.. u,. u,. because of the fairly simple composition of the amphi 78(4) 84(3) 80(3) -2(3) 14(3) -3(2) bole. TheA cation was removed from the refinement T(1) model, and a series of difference-Fourier maps were n21 66(3) 79(3} 67(3) 2(3) 14(2) -3(3) calculated through the A site (Fig. 1). The (100) M(1) 75(7) 69(6) 56(7) 0 20(6) 0 section at x = 0 shows significant displacement of M(2) 61(6) 70(6) 62(7) 0 19(4) 0 the maximum electron-density along the 2-fold axis M(3) 68(9) 57(9) 68(9) 0 7(6) 0
(Fig. la) parallel to [010], the A(2) position. The (010) M(4) 134(4) 110(4) 120(4) 0 67(3) 0 section at y = 1/z shows significant displacement of the 0(1) 81(8) 122(9) 89(9) ·13(7) 24(7) -13(7) maximum electron-density within the mirror plane 90(8) 7(7) (Fig. lb), the A(m) site. In order to assess the relative 0(2) 83(8) 91(5) 12(7) 2(7) electron-densities at these different sites, a difference 0(3) 106(11) 110(10) 111(11) 0 40(9) 0 Fourier map was calculated in the plane of the maxi 0(4) 140(9) 94(8) 119(9) -3(7) 46(8) -26(7) mum electron-density through the central A-site 0(5) 111(9) 157(9) 107(9) 43(8) 13(7) -7(7) (Fig. lc). Note the variability in contour intervals in 0(6) 102(9) 158(10) 136(10) -52(8) 28(8) 13(7) thissection: above 6 e/A3, the interval is 0.1 e, where 0(7) 117(13) 107(12) 174(14) 0 56(11) 0 as below 6 e/A 3, itis 1.0 e; this was done to emphasize the relative electron-densities at the A(2) and A(m) positions, and to show that there is an electron-density minimum (dotted) at the central A(2/m) position. Figure lc indicates that there is very slightly less isotropic displacement factors were used for the A(m) electron-density at A(m) relative to A(2), but the and A(2) positions, and were fixed at 0.02 for each difference is very small, and thus the electron site. Full-matrix least-squares refmement converged to densities are approximately the same. an R index of 2.9% and an wR index of 3.0%. Final The coordinates of the A(m) and A(2) positions positional parameters and equivalent isotropic were taken from the difference-Fourier maps, and displacement parameters are given in Table 2, Na was initially split equally between the two sites; anisotropic displacement parameters in Table 3 and 24 THE CANADIAN MINERALOGIST
TABLE 4. SELECTED INTeRATOMIC DISTANct!S (.&,l AND ANGLES (0) TABLE 4. (cont.) FOR FLUOR-EDEN(TE --- n2l tetrahedron nll-0(1) 1.644(2) Ml2l-0(1} x2 2.097(2) 0(2)-0(4) 2.763(3) o<2l-n2l-014l 117.3(1) nll-0(5) 1.673(3) · M(2)-0(2)b x2 2.062(2) _ 0(2)-0(6) 2.670(3) 0121-n21-0I6l 108.6(1) nll-0(6) 1.670(2) M(2)-0(4lc x2 1.998(2) 0(2)-0(5)a 2.678(3) 0(2)-n2l-0(6)a 109.6(1) nll-0(7) �
Alm)-0(5) x2 3.020(7) Al2l-0(5l x2 2.687(7) 0(1)-0(2)b x2 3.019(3) 011l-MI2l-012lb x2 93.1(1)
A(m)-0(6) x2 3.131(6) A12l-0(6) x2 · 3.382(8) 0(1)-0(2)d x2 2.780(3) 011l-MI2l-012ld x2 83.8(1) A(m)-0(6) x2 2.672(6) A(2)-0(6) x2 2.779(6) 0(1)-0(4)c x2 2.946(3) 011 l-M(2)-0(4)c x2 92.011) Alm)-016) x2 3.479(7) Al2l-016l x2 3.354(7) 0(2)d-014lc x2 2.941(3) 0(2)d-MI2l-0(4)c x2 92.8(1)
A(m)-0(7) 2.375(11) A12l-017l x2 • 2.415(4) 012ld-t114lf x2 2.869(3) 012ld-MI2l-014lf x2 89.9(1) Alml-0(7) 2.509(12) A (2)-0(7) x2 . 3.473(6) 0(1)-0(1)e 2.702(4) 011 l-MI2l-011 le 80.211) A(m)-0(7) 3.204(10) 0(4)c-014lf 2.982(4) 0(4)c-MI2l-014lf 96.5(1) Alml-0(7) 4.244(11) <0-0> � <0-M(2)-0> �
n 1) tetrahedron M(3) octahedron
0(1)-0(5) 2.759(3) 0(1)-T(1)-0(5) 112.6(1) 0(1)-0(3)a x4 3.070(3) 011 l-M(3)-0(3)a x4 97.4(1) 0(1)-0(6) 2.735(3) 0(1)-T(1)-0(6) 111.3(1) 0(1)-0(3)e x4 2.696(3) 011l-Mi3)-0(3)e x4 82.6(1) 0(1)-0(7) 2.734(3) 0(1)-T(1)-0(7) 111.8(1) 011l·OI1lg x2 2.702(4) 0(1)-M(3)·0(1)g x2 82.4(1) 0(5)-0(6) 2.669(3) Ol5l-n1l-OI6l 106.0(1) 0(1)-011lh x2 3.084(4) 0(1)-M(3)-0(1 )h x2 97.6(1) 0(5)-0(7) 2.654(3) 0(5)-nl)-0(7) 105.6(2)
0(2)-0(2)e 2.910(4) 012l-MI4l-0(2)e 74.3(1) 012l-014lc x2 3.121(3) Ol2l-MI4l-014lc x2 82.6(1) selected bond-distances, angles and polyhedral edge- 0(2)-0(4)i x2 2.941(3) 0(2)-Mi4)-0(4)i x2 76.9(1) lengths are given in Table 4. Observed and calculated 0(2)-0(5)c x2 3.494(3) 012l-MI4l-016lc x2 87.4(1) structure-factors are available from the Depository of 0(4)c-0(5)i x2 3.352(3) 0(4)c-M(4)-0(5)i x2 84.7(1) Unpublished Data, CISTI, National Research Council 0(4)c-0(6)c x2 2.674(3) 0(4)c-MI4)-0(6lc x2 63.3(1) of Canada, Ottawa, Ontario KIA OS2. 0(5)c-0(6)c x2 2.669(3) 0(6)c-M(4)-0(6)c x2 61.6(1) 0(5)c-0(6)j x2 2.985(3) 0(6)c-M(4)-0(6)j x2 69.9(1) 0(6)c-0(6)j 3.493(6) 0(6)c-MI4l-016lj 85.6(1) Electron-microprobe analysis
Following collection of the X-ray intensity data, the a= x, y, z+1; b = x, y, z-1; c = Y..-x, Y..-y, 1-z; d = -x, y, 1-z; crystal was mounted in �poxy, polished, carbon-coated e= -x, y, 1-z; f = x-Y.., Y.o-y, z-1; g = -x, y, -z; h= x, -y, z; i = x-%, *-y, z; j = x+ %, y-%, z and analyzed by electron microprobe. The analyses were done on a fully automated Cameca SX-50 according to the procedure of Raudsepp et al. (1991). As a control on accuracy, diopside and forsterite of in Table 5. As the chemical composition of this crystal known compositions were analyzed at the same time; is a very important factor in this work, we repeated the these showed close agreement with the nominal com- electron-microprobe analysis (a different set of positions. Ten points were analyzed on the crystal; the 10 points) several months later, and bracketed the results of individual analyses are given in Table 5, as analysis of fluor-edenite by analysis of a sample of significant zoning is present in the crystal. The aver- olivine and one of diopside of known compositions. age composition was calculated from the distribution The latter compositions determined here were found to of compositions over the crystal, and this also is given be accurate, supporting the results for fluor-edenite; A-SITE DISORDER IN FLUOR-EDENITE 2 5
'TABLE 5. RESULTS OF ELECTRON-MICROPROBE ANALYSIS EDENITE PARGASITE OF SYNTHETIC 'FLUOR-EDENITE'
End- Min.• Max.• Mean• Mean' member
Si02 48.78 44.67 46.76 47.22 60.16
Al203 8.98 13.36 11.32 11.00 6.08
MgO 24.04 22.87 23.23 23.16 24.06
CeO 12.46 13.11 12.92 13.03 13.38
Na20 3.58 3.17 3.38 3.41 3.70
F 6.14 6.23 6.08 5.01 4.63
OaF -2.16 -2.20 -2.14 -2.11 -1.91
Total** 100.81 100.10 100.56 100.72 100.00
Si 6.766 6.225 6.507 6.555 7.000
AI 1.246 1.775 1.493 1.445 1.000
IT 8.000 8.000 8.000 8.000 8.000 TREMOLITE HORNBLENDE AI 0.221 0.432 0.364 0.355 (6�1
Mg 4.963 4.783 4.819 4.793 5.000
IC 5.184 5.215 6.183 5.148 5.000 FIG. 2. Variation in composition with regard toA-site Na and [6] -coordinated AI across the crystal used for the collec
Mg 0.208 0.215 0.183 0.148 tion of X-ray intensity data.
Ce 1.849 1.970 1.926 1.938 2.000
2.057 2.185 2.109 2.086 2.000
0.961 0.862 0.912 0.918 1.000 Na indicates that the composition of the stable amphibole
• from first set of analytical results; in this system is very sensitive to the temperature t from second set of analytical results; of crystallization. • • the totals are a trifle high; using the ideal F-content reduces The site populations in this crystal are of particular tham by -0.5 wt%. interest. Firstly, the composition is simpler than most natural amphiboles, and there should be close agree ment with established site-population and stereo the latter are similar to the values obtained previously chemical relations. Secondly, this composition (Table 5), and also suggested that the mean compo provides an opportunity to examine A-cation posi sition is fairly representative of the whole crystal. tional disorder (Hawthorne & Grundy 1972) for a Unit formulae were calculated on the basis of 23 simple A-site chemistry with [4lAJ present and Fat the ' equivalent oxygen atoms. The analyzed F values 0(3) position. (-5.1 wt%)are slightly higher than the nominal value The C- and T-group cations'are Mg;Aland Si, with (4.5 wt%); as the structure cannot accommodate Fin atomic numbers of 12, 13 and 14, respectively. These excess of the ideal amount, the analyzed values are scatter X rays in a very similar fashion, and hence we thus a little high (albeit equal to the nominal value cannot use site-scattering refinement to derive site within two standard deviations). populations. However, thesB· cations are of very differ
ent sizes ([6lMg :::: 0.72 , [6lAl :::: 0.535, [4lAl = 0.39, DISCUSSION [4lSi :::: 0.26 A; Shannon 1976) and thus the mean bond-lengths can be used to derive the site popula Although the ideal composition of the synthesized tions. crystal is end-member fluor-edenite, both the electron microprobe analyses and the X-ray data show that it T(1) andT(2) sites deviates significantly from this composition. The com positional variation in the fragment used for the X-ray For an amphibole of this composition, [4lAl work (Fig. 2) lies within the compositional field of is expected to order at T(l). In synthetic fluor edenite. The spatial distribution of compositions amphiboles with no [4lAl,
TABLE 6. SELECTED MEAN BOND-LENGTHS CAl IN SYNTHETIC cation radius. A value of 2.055 A is obtained, in good FLUOR-AMPHIBOLES agreement with the observed value of 2.053 A. Ruor- Ruor- K-fluor- Ruor- richterito tromollte richterite edenite Ungaretti et al. (1981) and Hawthorne (1983) have published "ideal"
n21 Si Si Si Si bond-lengths. These procedures give M(2) occu
M(2) Mg Mg Mg (Mg,AI) M(4) site
Ref. (1) (2) (1) (3) structure-refmementhas shown the presence of signif icant electron-density at M(4'), a site very close to Refs: (1) Cameron et al. (1983); (2) Cameron & Gibbs (1973); M(4) but displaced along the 2-fold axis toward the (3) this study. octahedral strip (Oberti et al. 1992). Difference Fourier maps of fluor-edenite, calculated with the M(4) cation removed from themodel, show no sign of an M(4') site, and neither didanalogous Fourier maps. Bocchio et al. (1978) both gave predictive curves for An M(4') site was inserted into the refmement proce the amount of l4lAI in amphibole based on
Results for fluor-edenite the A cavity is surrounded by eight equivalent T(l) 1etrahedra, these tetrahedra are not all equivalent The difference- Fourier maps (Fig. 1) show that through the point-symmetry operations at the A(2/m) A(m) and A(2) are both occupied; there is a minimum position. This is illustrated diagrammatically in at the A(2/m) position and a saddle-point at the A(1) Figure 3. Consider first the A site on the left of the position. Thus the A(l) position is not occupied within figure. The adjoining T(l) sites fall into two sets, T(l)' the resolution of the X-ray method. Furthermore, site and T(l)", that are distinct with regard to the point scattering refinement shows Na (+ minor Ca) to be symmetry at the A(2/m) position. Thus when consi equally partitioned between the A(2) and A(m) sites. dering energy calculations based on a single cluster Firstly, we can say that the ordering is not the result of centered on one A(2/m) site, it is necessary to treat (OH,F) substitution at 0(3) or (Ca,Na) substitution at substitutions at the T( I)' and T( I)" sites as distinct. M(4), as these do not occur for this specific compo Inspection of Table 4 shows that A(2) has short sition. Similarly, the suggestion of Hawthorne bonds to 0(5) and 0(7), whereas A(m) has short bonds & Grundy (1978) that K orders at A(m) and Na orders to 0(6). Adjacent A cavities are connected, and each at A(2) is not correct. From the above discussion of T(l) tetrahedron is adjacent to two cavities (Fig. 3). site populations in the structure, the only possible This immediately suggests a reason for the equally factors affecting the local arrangement of charges are occupied A(2) and A(m) sites in the fluor-edenite (Al,Si) disorder at T(l) and (Al,Mg) disorder at M(2). crystal examined here. If a specific T(l) site is occu Docka et al. (198 7) found the minimum-energy pied by Al, the adjacent 0(5), 0(6) and 0(7) anions position for the case of Al occupancy of T(1) at the have a significant bond-valence deficiency. In one A(l) position, with an A(l)-A(l) displacement of adjacent A-cavity, Na occupies the A(m) position and -0.6 A across the 2-fold axis. This is not compatible increases the bond-valence incident to the associated with the results for this fluor-edenite (Fig. 1, Table 2). 0(6) anion. In the other adjacent A-cavity, Na occu Either the small amount of l6lAl has drastically altered pies the A(2) position and increases the bond-valence the general energetic situation here, or the omission incident to the associated 0(5) and 0(7) anions. Each of local relaxations associated�with the different bridging anion coordinated to T(l) thus has additional arrangements of charges has adversely affected the incident bond-valence from this local arrangement, calculations. and will cause equal occupancy of the A(m) and A(2) It is difficult to reconcile the (e xactly) equal parti positions. This arrangement may propagate along the tioning of Na between the A(2) and A(m) sites with the chain of A cavities parallel to the c axis to form an intermediate (i.e., non-end-member) composition of ordered arrangement, but there is no way that this the crystal. In fact, it seems reasonable to conclude arrangement can be propagated to adjacent chains of that the ordering must be fairly independent of small A cavities, and thus long-range C2/m symmetry with changes in bulk composition in this particular case. the usual cell of amphibole is preserved. The key feature here is the recognition that although The argument advanced above for equal occupancy
FIG. 3. The A-site cavity and surrounding anions viewed down [010]. The T(l)' and T(l)" tetrahedra formtwo sets that are notrelated by the point sy=etryat the A position;a single T(l) tetrahedron has different relationships to the central A-position in each cavity. The broken lines show the limits of each A cavity. A-SITE DISORDER IN FLUOR-EDENITE 29
of the A(m) and A(2) positions is general and suggests CRoMER, D.T. & LmERMAN, D. (1970): Relativistic calcula that such a configuration should be the usual case in tion of anomalous scattering factors for X rays. J. Chern. Phys. 53, 1891-1898. amphiboles. However, the experimental sit�ation is less clear, as the assignment ofA-site populations is -- MANN, B. (1968): commonly complicated by the presenc� of :K,,.and the & J. X-ray scattering factors com puted from numerical Hartree-Fock wave functions. argument may need to. be modified if other orderini Acta. Crystallogr.A24, 321-324.' schemes [e.g., AI at M(2), Na at M(4), OH at 0(3)] also are present; these factors are··currently being DOCKA, J.A., POST, J.E., BISH, D.L. & BURNHAM, C.W. et al. examined in detail by Oberti (work in progress). (1987): Positional disorder of the A-site cations in C2Jm amphiboles. Model energy calculations and probability ' ' , · CONCLUSIONS studies. Am.Mineral. 72, 949-958, c, ;:}' 1, The compositional results suggest that the devia- GmBs, G.V. (1966): Untitled article. InAGI Short Course tion of amphibole composition from nominal in the Lecture Notes on Chain Silicates, 1-23. fluor-edenite system is a function of the temperature of crystallization. -- & PREwrrr, C.T. (1968): Amphibole cation site disor 2. Tetrahedrally coordinated AI is ordered at T(l) in der. Int. Mineral. Assoc., Proc., Fifth Gen. Meet. this synthetic fluor-edenite. (Cambridge, 1966). (abstr.). 3. Octahedrally coordinated AI is ordered at M(2). 4. ANa is equally partitioned between the A(m) and GRAHAM, C.M. & NAVROTSKY, A. (1986): Thermochemistry A(2) sites in this synthetic fluor-edenite. of the tremolite - edenite amphiboles using fluorine ana 5. The equal occupancy of theA(m) andA(2) sites by logues, and applications to amphibole - plagioclase - quartz equilibria, Contrib. Mineral. Petrol. 93, 18-32. Na can be explained on the basis of local bond valence requirements of the anions coordinating the GREENWOOD, H .J. (1979): Thermodynamic properties of T(l) site. edenite. Geol. Surv. Can., Pap. 79-lB, 365-370.
ACKNO�GEMENTS HAWTHORNE, F,C, (1983): The crystal chemistry of the amphiboles. Can. Mineral. 21, 173-480. We thank Luciano Ungaretti and an anonymous
reviewer for comments that materially improved this __ & GRUNDY, H.D. (1972): Positional disorder in the paper. Financial support was provided by the Natural A-site of clino-amphiboles. Nature Phys. Sci. 235, 72-73. Sciences and Engineering Research Council of Canada
via a Post-graduate Fellowship to PCB and Operating, -- & __ (1973a): The crystal chemistry of the amphi Equipment and Infrastructure Grants to FCH. boles. I. Refinement of the crystal structure of ferro tschermakite. Mineral. Mag. 39, 36-48.
REFERENCES -- & __ (1973b) : Thecrysta l chemistry of the amphi boles. II. Refinement of the crystal stmcture of oxy-kaer BoCCHIO, R., UNGAREITI, L. & Rossi, G. (1978): Crystal sutite. Mineral. Mag. 39, 390-400. chemical study of eclogitic amphiboles from Alpe Arami, Lepontine Alps, southern Switzerland. Soc. !tal. Mineral. -- & __ (1977): The crystal chemistry of the amphi Petrol. Rend. 34, 453-470. boles. ill. Refmement of the crystal stmcture of a sub silicic hastingsite.Mineral. Mag. 41, 43-50.
BROWN, I.D. (1981): The bond-valence method: an empirical -- & -- (1978): The crystal chemistry of the amphi approach to chemical stmcture and bonding. In Stmcture boles. VII. The crystal stmcture and site chemistry of and Bonding in Crystals II (M. O'Keeffe & A. potassian ferri-tatamite. Can. Mineral. 16; 53-62. Navrotsky, eds.). Academic Press, New York (1-30).
HERITSCH, H. (1955): Bemerkungen zurBchreibung der SHANNON, -- & R.D. (1973): Empirical bond-strenth Tschermak:S A29, kristallchemischen Formel der Hornblende. bond-length curves for oxides. Acta Crystallogr. Mineral. Petro gr. Mitt. 5, 242-245. 266-282.
HINRICHSEN, T. & SCH0RMANN, K. (1977): Experimental CAMERON, M. Gmas, G.V. & (1973): The crystal stmcture investigations on the Na/K substitution in edenites and and bonding of fluor-tremolite: a comparison with pargasites. Neues Jahrb. Mineral. Abh. 130, 12-18. hydroxyl ttemolite.Am. Mineral. 58, 879-888. NA, K.C., McCAULEY, M.L., CRISP, J.A. & ERNST, W.G. -----, SUENO, S., PAPIKE, J.J. & l'REwrrr, C.T. (1983): High (1986): Phase relations to 3 kbar in the systems edenite + temperature crystal chemistry of K and Na fluor H20 and edenite + excess quartz + H20. Lithos 19, richterites.Am. Mineral. 68, 924-943. 153-163. 30 THE CANADIAN MINERALOGIST
OBERT!, R., UNGARE'ITI, L., CANNILLO, E. & HAWTHORNE, UNGARE'ITI, L. (1980): Recent developments in X-ray single F.C. (1992): Thebehaviour of Ti in amphiboles. I. Four crystal diffractometry applied to the crystal-chemical and six-coordinated Ti in richterite. Eur. J. Mineral. 4, study of amphiboles. God. Jugosl. Cent. Kristalogr. 15, 425-439. 29-65.
PAPIKE, J.J., Ross, M. & CLARK, J.R. (1969): Crystal -----. SMITH, D.C. & Rossr, G. (1981): Crystal-chemistry chemical characterization of clinoamphiboles based on by X-ray structure refinement and electron microprobe five new structure refinements.Mineral. Soc. Am., Spec. analysis of a series of sodic--calcic to alkali amphiboles Pap. 2, 117-136. from the Nyoo eclogite pod, Norway. BulL MineraL 104, 400-412. PREWITT, C.T. (1963): Crystal structures of two synthetic amphiboles. Geol. Soc. Am., Spec. Pap. 76, 132-133 WIDMARK, E.T. (1974): An edenite-forming reaction: (abstr.). hydrothermal experiments. Neues Jahrb. Mineral., Monatsh., 323-329. RAUDSEPP, M., 'fuRNOCK, A.C. & HAWTHORNE, F.C. (1991): Amphibole synthesis at low pressure: what grows and what doesn't. Eur. J. Mineral. 3, 983-1004.
SHANNON, R.D. (1976): Revised effective ionic radii and systematic studies of interatomic distances in halides and Received April 24, 1992, revised manuscript accepted April chalcogenides.Acta Crystallogr.A32, 751-767. 27,1993.